A fast and scalable ensemble of global models with long memory and data partitioning for the M5 forecasting competition

This work presents key insights on the model development strategies used in our cross-learning-based retail demand forecast framework. The proposed framework outperforms state-of-the-art univariate models in the time series forecasting literature. It has achieved 17th position in the accuracy track of the M5 forecasting competition, which is among the top 1% of solutions.     

Combining forecasts for universally optimal performance

There are two potential directions of forecast combination: combining for adaptation and combining for improvement. The former direction targets the performance of the best forecaster, while the latter attempts to combine forecasts to improve on the best forecaster. It is often useful to infer which goal is more appropriate so that a suitable combination method may be used. This paper proposes an AI-AFTER approach that can not only determine the appropriate goal of forecast combination but also intelligently combine the forecasts to automatically achieve the proper goal. As a result of this approach, the combined forecasts from AI-AFTER perform well universally in both adaptation and improvement scenarios. The proposed forecasting approach is implemented in our R package AIafter, which is available at https://github.com/weiqian1/AIafter.     

A hybrid machine learning model for forecasting a billing period’s peak electric load days

Many models have been studied for forecasting the peak electric load, but studies focusing on forecasting peak electric load days for a billing period are scarce. This focus is highly relevant to consumers, as their electricity costs are determined based not only on total consumption, but also on the peak load required during a period. Forecasting these peak days accurately allows demand response actions to be planned and executed efficiently in order to mitigate these peaks and their associated costs. We propose a hybrid model based on ARIMA, logistic regression and artificial neural networks models. This hybrid model evaluates the individual results of these statistical and machine learning models in order to forecast whether a given day will be a peak load day for the billing period. The proposed model predicted 70% (40/57) of actual peak load days accurately and revealed potential savings of approximately USD $80,000 for an American university during a one-year testing period.     

Fiscal rules to tame the political budget cycle: Evidence from Italian municipalities

The paper provides evidence that fiscal rules can limit the political budget cycle. It uses data on Italian municipalities during the 2000s and shows that: 1) municipalities are subject to political budget cycles in capital spending; 2) the Italian sub-national fiscal rule (Domestic Stability Pact, DSP) introduced in 1999 has been enforced by the central government; 3) municipalities subject to the fiscal rule show more limited political budget cycles than municipalities not subject to the rule. In order to identify the effect, we rely on the fact that the domestic fiscal rule does not apply to municipalities below 5000 inhabitants. We find that the political budget cycle increases real capital spending by about 10–20 percent on average in the years prior to municipal elections and that municipalities subject to the DSP show a pre-electoral increase in capital spending which is only a quarter of the one of municipalities not subject to the rule.     

Relative age effects in political selection

We exploit a regression discontinuity design to provide causal evidence of the relative age effect (RAE) on a long-run adult age outcome: Political selection. We find strong evidence of the RAE in politics in Finland. However, the effect is heterogeneous: We find that male candidates born early in the calendar year have a significantly higher probability of getting elected to the parliament but no similar RAE applies to female candidates nor to municipal elections. Moreover, this effect only takes place in the most competitive parliamentary districts and is present only for some parties. We also find that in all the groups where the RAE does not exist, early-born candidates are under-represented suggesting attrition of talent in the candidate placement. Overall, our results show that seemingly artificial cutoffs imposed by the government have persistent consequences even on the selection to the highest positions of power within a society.     

The non-linear impact of oil price on the oil demand

ABSTRACT

A panel smooth transition regression model was adopted to analyse the non-linear impact of oil prices on oil demand. Data for 42 countries was obtained from the International Energy Agency for the time period spanning from January 1990 to June 2017. The results indicate that a threshold value does exist. Furthermore, when the oil price was lower than this threshold value, a positive relationship between oil price and oil demand was observed. When the price of oil was higher than the threshold value, however, a negative relationship between price and demand was found.     

Modelling demand for lotto using a novel method of correcting for endogeneity

Modelling lottery sales as a function of the mean, standard deviation and skewness of the probability distribution of returns potentially gives insights into how the design of a game could be modified to maximise net revenue. But use of OLS is problematic because the level of sales itself affects values of the moments (and insufficient instruments are available for IV regression). We draw on the concept of a rational expectations equilibrium, developing a new regression model which corrects for endogeneity where the causal impact of the dependent variable on the right-hand side variables is deterministic. We apply the model to data on lotto sales from Spain. Using the Spanish data, we show that results provide more reliable guidance to lottery agencies because accounting for endogeneity leads to significantly different results from OLS and these results have superior performance in out-of-sample forecasting of sales. More generally, results prove consistent with the Friedman-Savage explanation of why people buy lottery tickets and with evidence from racetrack data that ‘bettors love skewness’.     

Determining the predictors for ease of sleep while on aircraft: Regression and qualitative analyses

BackgroundFactors predicting passengers’ ability to fall asleep and levels of sleep anxiety, while traveling on a commercial flight, are investigated through a two-study mixed design.MethodsData collected from approximately 400 participants contributed to the development and validation of multiple regression equations and model fit analysis; and participants responded to related open-ended questions.ResultsRegression equations yielded between two to seven predictors and predicted between 6.7% and 27.7% of the variance, ps < .001. Model fit was strong in all cases. An inductive qualitative approach provided detailed insight into passengers’ concerns and barriers over sleeping on a commercial flight.DiscussionAs the field of commercial aviation continues growing, researching and understanding passengers' experiences and perceptions is crucial to the success of the industry as consumers ultimately drive the market. Passengers’ perceptions of sleep quality on commercial aircraft is a key factor influencing their traveling decision. Therefore, a better understanding of this phenomenon can provide crucial information to future passengers, airline companies, regulatory agencies, and manufacturers, potentially influencing the future success of the aviation industry.